• No results found

The impact of investor protection on the relationship between abnormal audit fees and audit quality

N/A
N/A
Protected

Academic year: 2021

Share "The impact of investor protection on the relationship between abnormal audit fees and audit quality"

Copied!
42
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst

(1)

The Impact of Investor Protection on the

relationship between Abnormal Audit Fees and

Audit Quality

Master Thesis

MSc Accountancy & Control, specialization Accountancy

Faculty of Economics and Business, University of Amsterdam

Name:

Marieke Smeets

Student number:

10456384

(2)

Statement of Originality

This document is written by student Marieke Smeets declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

Abstract

The purpose of this study is to examine the impact of investor protection on the relationship between abnormal audit fees and audit quality. Reynolds & Francis (2000) argue that audit quality is a tradeoff between the economic benefits of a client and the expected costs in case of audit failure. Based on prior research I argue that the level of investor protection rules out the economic benefits of a client. To examine this hypothesized relation, earnings management is used to proxy audit quality and is estimated by the Modified Jones Model. The level of investor protection is scored according to La Porta et al. (1998).

The findings of this study indicate that there is a difference in effect between positive abnormal audit fees and two forms of earnings management. In case of managing earnings upwards, results show that positive abnormal audit fees ensure an increase in positive discretionary accruals. This result is supported by the economic dependence theory of DeAngelo (1981a). However, positive abnormal audit fees mitigate the level of discretionary accruals in case of income decreasing earnings management. This finding is in accordance with the effort theory of Hay et al. (2006). Additionally, the results argue that negative abnormal audit fees do not influence the level of earnings management and thus audit quality. Contrary to the expectations, the results show no evidence for a significant relationship between the level of investor protection of a country and the level of earnings management. This can be interpreted as investor protection standalone does not influence earnings management. The results of this study only find evidence for the mitigating effect of investor protection on the relationship between positive abnormal audit fees and income increasing earnings management.

(4)

Table of Contents

1 Introduction ... 5

2 Literature Review & Hypotheses Development ... 8

2.1 Audit Quality ... 8

2.1.1 Definitions ... 8

2.1.2 Auditor Independence ... 8

2.2

Audit Fees & Audit Quality ... 9

2.2.1 Economic Dependence Theory ... 9

2.2.2 Effort Theory ... 9

2.2.3 Abnormal Audit fees & Audit Quality ... 9

2.3 Investor protection ... 12

3 Research design and methodology ... 14

3.1 Measuring abnormal audit fees ... 14

3.1.1 Client attributes to audit fees ... 14

3.1.2 Auditor’s attributes to audit fees ... 15

3.1.3 The audit fee model ... 16

3.2 Measuring earnings management ... 16

3.3 Measuring investor protection ... 18

3.4 Data and sample selection ... 19

3.5 Empirical model ... 19

4 Analyses and Results ... 21

4.1 Descriptive statistics ... 21

4.2

Correlation ... 22

4.3 Results of the estimation of abnormal audit fees ... 23

4.4 Main analysis ... 24

4.5 Additional analysis ... 28

5 Conclusion ... 34

5.1 Summary ... 34

5.2 Conclusion ... 35

5.3 Limitations and future research ... 35

6 Reference list ... 37

(5)

1

Introduction

The public trust in auditor independence was being questioned after the Enron Scandal came to light. Enron misstated the amount of revenue and assets, and presented non existing accounts in their financial statements. At that time, Enron was a client of the Houston office of Arthur Andersen. The auditors of the Houston office were aware that these false financial statements were misrepresentative. Enron was the largest client that the Arthur Anderson Houston-office served. According to Schmidt (2002) there is evidence that partners in the Houston-office were overriding quality control standers, which were set by Arthur Anderson’s head-office. This could be a signal of higher tolerance of aggressive accounting by their clients (Krishnan, 2005). After Enron and several other accounting scandals in the beginning of this century, the need for an auditor to act independent in fact as well as in appearance increased. When an auditor is perceived to be independent by investors, the perceived quality of the financial statements increases (Moore et al, 2006). According to the IASB (2014), the main objective of financial reporting is to provide reliable information about the reporting entity, that will be useful for decision-making for capital providers. Therefore, the role of the auditor is to provide assurance to those capital providers and other stakeholders.

There exists a vast amount of literature regarding auditor independence. Especially about economic factors such as auditor tenure, auditor size, audit fees and client importance (Frankel et al., 2002; Reynolds & Francis, 2000; Antle et al., 2012; Chung & Kallapur, 2003). All these factors relate to the economic dependence of an auditor towards their client. According to DeAngelo (1981a), when an auditor receives relatively high audit fees or provides multiple services to the client, the auditor may be more depending on the client. According to the Simunic (1984), an economic bond with a client could incentivize auditors to report more favorable to important clients for the firm. It is probable that the level of economic dependence in the Enron scandal had led to the misrepresented financial statements, since Enron was the largest client of the Houston-office.

Therefore, multiple studies focus on economic dependence and the impact on audit quality. Antle et al. (2012) examine the height of audit fees regarding audit quality, and they found that auditor independence comprises with respect to greater audit fees. Likewise, Frankel et al. (2002) show that auditor independence is impaired by an increased ratio of non-audit fees.

(6)

theory, Reynolds & Francis (2000) and Chung & Kallapur (2003) find that auditors act more conservative when there is increased client importance.

For this study, abnormal audit fees will be used as a proxy for economic dependence. Abnormal audit fees are the excessive part of remuneration paid to the auditor. Hoitash et al. (2007), Blankley et al. (2012) and Choi et al. (2010) all examined the relationship between abnormal audit fees and audit quality. However, all studies provide ambiguous results. Therefore, the question still rises whether abnormal audit fees compromises auditor independence, or whether it improves audit quality by an auditor exerting more effort. According to Reynolds & Francis (2000), audit quality will be the result of the tradeoff between economic dependence and the damage caused by the legal environment in case of audit failure. Francis & Wang (2008) conclude that conservatism is higher in countries with greater investor protection. This outcome supports the theory that the legal environment in which an auditor operates will affect the relationship between client importance and audit quality. In most countries corporate governance and the audit profession is already being regulated, attempting to limit the level of economic dependence, which could be a threat to auditor independence. In the United States, this has been done via the implementation of the Sarbanes Oxley Act in 2002. This act tends to restore the public trust in the accounting profession, by protecting investors by providing regulations to improve the reliability of financial information.

Overall, the US has the highest scores regarding the investor protection according to La Porta (1998). The level of investor protection in the US might be an explanation why the outcomes differ from the primary expectations. Most of the previous studies regarding abnormal audit fees and audit quality were focused on the United States. Therefore, the relationship between abnormal audit fees and audit quality might be influenced by the level of investor protection. That is why this thesis will attempt to answer the following question: “Does investor protection mitigate the relation between abnormal audit fees and audit quality?” By examining this question, this research can contribute to the literature in two ways. First, it could provide insights into the nature of the relationship between abnormal audit fees and audit quality. Secondly, the ambiguous results of prior research about the relationship between economic dependence and audit quality can be caused by the legal environment of a country. Most of prior research made use of US data, the US had the highest investor protection

(7)

according to LaPorta (1998). But does the relationship between increased economic dependence and audit quality still holds when investor protection is low?

It is important to see what the impact is of investor protection on the relationship between client importance and audit quality, because it increases the relevance of audited financial statements for its users. When audit quality is perceived as high, the legal environment a firm operates in, can still be a weakness with regard to the reliability of the audited financial statements. So in this research I will examine whether investor protection is of greater importance than economic dependence, to increase audit quality.

This study will capture economic dependence by abnormal audit fees and audit quality will be proxied by earnings management. Based on the Modified Jones Model the discretionary accruals will be estimated. Subsequently, the abnormal audit fees will be estimated as the error terms of the expected audit fee model. The level of investor protection in a country will be scored using the scores of La Porta (1998). To test the hypothesized relationships two regressions will be carried out, one over the total sample and the other distinguishes positive and negative abnormal audit fees. Finally, an additional analysis will be performed over two subsamples of discretionary accruals.

The findings of this study indicate that there is a difference in effect between positive abnormal audit fees and two forms of earnings management. No relation is found between negative abnormal audit fees and earnings management. Furthermore, the level of investor protection does not seem to influence the level of earnings management. Some evidence is found that indicates the mitigating effect of investor protection on the relationship between abnormal audit fees and earnings management as a proxy for audit quality.

The remainder of this thesis is organized as follow. In paragraph two, an overview of the literature regarding auditor independence and audit quality will be provided. Different aspects that could potentially influence those subjects will be identified. In the third paragraph, the research method will be discussed. Subsequently, the results and analyses will be provided in the fourth paragraph. In the final paragraph, a summary and conclusion of this research will be provided.

(8)

2

Literature Review & Hypotheses Development

In the following paragraph the key concepts around auditor independence, audit quality and investor protection will be discussed. Also a review of the existing literature around these key concepts will be provided. Subsequently, the development of the hypotheses will be discussed.

2.1 Audit Quality

2.1.1 Definitions

“The quality of audit services is defined to be the market-assessed joint probability that a given auditor will both (a) discover a breach in the client's accounting system, and (b) report the breach” (DeAngelo, 1981a, p. 186). The detection of material errors depends on the knowledge and skills of the auditor and the way these statements are audited (Knechel et al, 2013). An error is material when it influences economic decisions made by the financial statement users. Palmrose (1988) defines the term audit quality as the risk that a financial statement contains no inaccuracies. According to Palmrose (1988) this means that the greater the chance that the audited statements do not contain any inaccuracies, the higher the quality of the audited statements is. The definitions of audit quality by DeAngelo (1981a) and Palmrose (1988) mention that the quality of the audit will be determined by the detection of material errors, if present, in the audited financial statements. But due to the complexity to determine if there were any material errors in the financial statements, there are multiple indirect methods developed to determine the quality of the audit output.

2.1.2 Auditor Independence

Besides the detection of material errors, it is also of interest that these material errors will be reported by the auditor. The chance that the auditor will report a breach is related to the level of independence of the auditor (DeAngelo, 1981a). Moore et al. (2006) argue that conflicts of interest impair auditor independence. Therefore, it is of great importance that an auditor is independent from his client, because he serves the public interest.

According to the rules of the Securities and Exchange Commission, hereafter SEC, an auditor is required to be independent of mind, as well as in appearance. This means that an auditor needs to act with integrity and objectively in the audit process. It is hard to determine the level of auditor independence in fact, because it is not measurable and observable. Auditor

(9)

independence of mind will be influenced by the auditor’s capacity to keep professional skepticism, objectivity and integrity (Hayes et al., 2015).

For users of financial statements, the perceived audit quality increases by observing that an auditor is independent in appearance. Which means that the users of the audited financial statements perceive the auditor to be independent. Independence in appearance involves the avoidance of significant circumstances that a reasonable an informed third party, could conclude that the professional accountant's integrity, objectivity or professional skepticism are being threatened (Hayes et al., 2015). Independence in appearance increases the perceived reliability of financial statements towards its users.

2.2 Audit Fees & Audit Quality

2.2.1 Economic Dependence Theory

DeAngelo (1981a) examines that the level of independence will be determined by the economic bond of the auditor with regard to his client. If an auditor is more depending on a client it is expected that he would be less inclined to report these material errors than when an auditor is less dependent on a client. This theory that is also confirmed by Simunic (1984). He concludes that auditor independence comprises by increased economic dependence.

2.2.2 Effort Theory

Hay et al. (2006) review existing literature with regard to audit fees. However, they find that expected audit fees were determined by the ‘supply’ proxies, such as client size, client complexity and client risk. They argue that abnormal audit fees are determined by the level of extra effort that is expended by the auditor. This may say that the higher the audit fee, the greater the effort en thus the higher the level of audit quality. This theory of Hay et al. (2006) is in contradiction with the theory of economic dependence regarding its impact on audit quality.

2.2.3 Abnormal Audit fees & Audit Quality

Abnormal audit fees represent the overpayment or underpayment of audit services (Choi et al., 2010). Normal audit fees represent the size, complexity and riskiness of an audit. The residual

(10)

Choi et al. (2010) find that positive abnormal audit fees have significant negative influence on the level of discretionary accruals. In contradiction to a positive abnormal audit fees, Choi et al. (2010) did not find a significant relationship with negative abnormal audit fees and audit quality. Meaning that an increased economic bond would lead to a lack of auditor independence and a greater tolerance towards earnings management. However, exerting less effort would not reduce earnings quality, according to Choi et al. (2010).

The study of Asthana & Boone (2012) examine whether positive abnormal audit fees, as well as negative abnormal audit fees, comprises audit quality. For determining audit quality, they use discretionary accruals and the meeting or beating of analysts’ earnings forecasts as proxies. The results of this study show that audit quality comprises as the magnitude of the abnormal audit fees increases. This holds for positive, as well as for negative abnormal audit fees. This suggests that negative abnormal audit fees, due to client bargaining power, will decrease the exerted effort and thus decrease audit quality. Asthana & Boone (2012) also conclude that audit quality is less comprised in the post-SOX period, than in the pre-SOX period. Meaning that SOX reduces the risk that an auditor succumbs to client pressure.

However, the results of Blankley et al. (2012) show that positive abnormal audit fees are negatively related with future restatements, after controlling for internal control quality. Restatements would reflect low audit effort in the year’s prior periods to the restatement (Blankley et al., 2012). Audit fees were lower in periods prior to the restatements, which suggests that audit effort was low in periods before the restatements.

The results of Krauß et al. (2015) confirm the theoretical expectations of DeAngelo (1981a) and Simunic (1984). They examine the association between abnormal audit fees and audit quality in the German market, using discretionary accruals as a surrogate for audit quality. A negative relationship is found between positive abnormal audit fees and the level of discretionary accruals. Suggesting that audit quality is impaired with positive abnormal audit fees. However, they are not able to find significant results in the period 2008-2010, when regulatory strength was increased. For negative abnormal accruals Krauß et al. (2015) find an insignificant negative relationship with the level of discretionary accruals. Xie et al. (2010) find no significant association between abnormal audit fees and audit opinions in the Chinese market.

Eshleman & Guo (2014) examine the relationship of positive abnormal audit fees with the level of discretionary accruals, while considering managers’ incentives and ability to manage earnings to meet-or-beat analysts’ forecasts. Their results show a negative relationship.

(11)

Thus audit quality increases with an increase in the magnitude of abnormal audit fees. The study of Eshleman & Guo (2014) supports the effort theory.

Hoitash et al. (2007) find a significant positive relationship between abnormal audit fees and audit quality. Absolute discretionary accruals and the standard deviation of residuals related to current accruals of cash flows are used as surrogates of audit quality. Hoitash et al. (2007) note economic bonding as a determinant of auditor behavior. These outcomes support the economic dependence theory.

Higgs & Skantz (2006) study the impact of fee residuals on the earnings response coefficient, this is a proxy for earnings quality. Their findings show a negative relationship between the magnitude of fee residuals and earnings quality. Supporting the economic dependence theory.

The results of prior studies are ambiguous about the nature of the relationship between abnormal audit fees and audit quality. A part of the literature supports the audit effort theory and the other part relies on the theory of economic dependence. Because of the Enron scandal, where client pressure was a threat to auditor independence, it seems more probable that in fact the economic dependence theory will override the audit effort theory. The more economically dependent an auditor is of a particular client, the more favorable he will report for this client (DeAngelo, 1981; Simunic, 1984). Therefore, the expectation is that an auditor allows greater discretion with regard to accounting accruals, when there is a strong economic bond.

H1a: There is a positive relationship between positive abnormal audit fees and the level of discretionary accruals

For negative abnormal audit fees, it seems probable that the audit effort theory will hold. According to this theory, negative abnormal audit fees will lead to less audit effort. Otherwise the audit engagement would not be profitable for the audit firm. Thus, the expectation is that less audit effort will result in declining audit quality (Asthana & Boone, 2012)

H1b: There is a positive relationship between negative abnormal audit fees and the level of discretionary accruals

(12)

2.3 Investor protection

As earlier mentioned, audit quality will result from a tradeoff between the economic benefits from a client and the potential litigation and reputation costs in case of audit failure (Reynolds & Francis, 2000). The legal environment regarding audit environments is influenced by laws and regulations regarding auditors. Such as independence restrictions and auditor liability (Maijoor & Vanstraelen, 2006).

The Security and Exchange commission (SEC) identified that several incremental changes within the law increase auditor’s legal exposure. The study of Geiger et al. (2002) indicate that due to the reduction in liability exposure, Big 6 auditors act less conservative. They find that Big 6 auditors issue less going concern modified opinions tot financially distressed companies since the new reform act.

Francis & Wang (2008) examine the effect of investor protection on the relationship among firms audited by Big 4 auditors and earnings quality. The results of this study show a lower level of earnings management in countries with a strong legal environment for Big 4 auditors, than clients of non-Big 4 auditors. According to Francis & Wang (2008) Big 4 auditors experience a greater reputation risk, than non-Big 4 auditors and therefore report more conservative to their clients. In contrast, Maijoor and Vanstraelen (2006) did not find a difference between these types of auditors. They examined the relation between investor protection and level of earnings management in Europe. Their findings show that a stronger legal environment decreases the level of earnings management, but this relation was regardless of the type of auditor.

The studies discussed in this paragraph measure the impact of the level of investor protection on audit quality. Their results show that there exists a positive relationship between the level of investor protection and audit quality. Due to an increased litigation risk, auditors would be more incentivized to perform their work with more caution (Francis, 2004). On the contrary, when investor protection is low, audit quality is expected to be lower. In line with the argumentation and findings of Francis & Wang (2008) and Geiger et al. (2002), the following hypotheses is developed:

H2: The level of investor protection has a negative relationship with the level of discretionary accruals

(13)

Uncertainties regarding the ambiguous outcomes of prior studies that examine the relationship between abnormal audit fees and audit quality could be due to differences in institutional characteristics. Such as less restrictive auditor liabilities, different corporate governance requirements or the difference in disclosure requirements. These institutional characteristics indicate the level of investor protection in a country.

According to Reynolds & Francis (2000) audit quality will be the result of the tradeoff between the benefits of reporting favorably for a client and the damage that misreporting can cause in case of audit failure. Auditors could be incentivized to tolerate more earnings management to avoid the dismissal by clients. However, this incentive could be reduced by stricter investor protection regimes. Where it is more likely that misreporting is detected and liability exposure for auditors is increased (Francis & Wang, 2008). Thus, the increased risk of auditor liability could mitigate the incentive that auditors have to report favorably for their clients, because of the negative consequences of litigation and reputation costs that auditors face. The expectation is that the hypothesized relationship between abnormal audit fees and audit quality can be mitigated by the level of investor protection.

H3: The level of investor protection negatively affects the relationship between abnormal audit fees and the level of earnings management

(14)

3

Research design and methodology

The following section will describe the formation of the sample selection. Subsequently, the characteristics and the operationalization of those variables that are expected to influence audit fees and audit quality will be discussed. Furthermore, the design of the empirical model to test the formulated hypotheses.

Figure 3.1 Conceptual model

3.1 Measuring abnormal audit fees

According to Hay et al. (2006) audit fees consist of two parts. The first part is the normal/expected audit fee, which will be determined by supply and demand attributes. The other part will be determined by specific characteristics of the auditor-client relationship. The part of the fee based on the auditor-client relationship, is called the abnormal audit fee. The abnormal audit fee is calculated as the difference between the fee paid and the expected normal audit fee.

3.1.1 Client attributes to audit fees

In the research of Hay (2013), he analyzes and describes important client characteristics that attribute to audit fees.

Client size:

The greater the firm, the more transactions take place. The size of a client indicates the extent of work the auditor has to perform. Total assets and total revenues are common indicators of

Independent variable: Abnormal audit fee

Dependent variable: Discretionary accruals Moderating variable:

(15)

client size. According to Hay (2013) both indicators have a significant positive relation with audit fees. He suggests both are equally good indicators of client size. In this study I will use that natural logarithm of total assets. Because the natural logarithm minimizes the effect of greater values with regard to the effect of smaller values.

Complexity:

Hay (2013) finds that there is a positive relation between the audit fee and the complexity of a firm. The complexity of a firm increases the inherent risk. The more complex, the higher the charged fee will be, because the auditor has to evaluate more procedures and controls within the organization. Hay (2013) distinct multiple types of complexity, namely geographical complexity and organizational complexity. Geographical complexity will be operationalized by using the percentage of foreign revenue relative to total revenue (FOREIGN) as a variable. The inventory and receivables divided by total assets (INVREC) reflects the organizational complexity. Inventory and receivables are complex to value according to Simunic (1984).

Financial position:

Prior research finds a negative relation between the financial position of a firm and the

The financial position of a firm influences the control risk of the auditor negatively (Hay, 2013). Therefore, the expectation is that the normal audit fee increases. To estimate the financial performance of a firm, the return on assets (ROA) and the dummy variable LOSS will be introduced. The leverage ratio of the firm (LEV) and the current ratio (LIQ) are indicators for the financial position of a firm. Prior year’s type of opinion can also be an indicator for the control risk.

3.1.2 Auditor’s attributes to audit fees

Big 4 auditors will usually charge a higher fee, because of their reputation. Prior research indicate that Big 4 auditors deliver higher audit quality, because they are driven by reputation costs (Reynolds & Francis, 2000; Francis & Yu, 2009). This will be measured by the dummy variable BIG4. The score of one will be assigned when the auditor is a Big 4 audit firm, and the score will be 0 in case of a non-Big 4 auditor. It is expected that this variable will have a positive relationship with the height of the audit fee.

(16)

3.1.3 The audit fee model

All previous described attributes contribute to the expected normal audit fee. The difference between the expected normal audit fee and the charged audit fee is the abnormal part of the audit fee. This residual is presented in the model as an error term and will be used as a proxy for abnormal audit fees. The residual represents overpayment in case of a positive number and underpayment when the residual is negative.

The following model is will be used to calculate the audit fee:

!"_$%&& = )*+ ),!"-$ + ).%/0&12" + )31"40&5 + )60/$ + )7!&4 + )8!19 + ):!/;; + )<=124 + ),*/?1"1/" + @

Table 3.1 Descriptions of the expected audit fee model

Variable Description

LNTA Natural logarithm of total assets

FOREIGN Percentage of foreign revenue relative to total revenue of the current year INVREC Inventory plus receivables divided by total assets

ROA Net income divided by total assets LEV

LIQ

Total debt of the firm divided by the total equity Current assets divided by current liabilities

LOSS 1 if the firm reported a loss in the prior year, otherwise 0 BIG4 1 if the firm has a Big 4 auditor, otherwise 0

OPINION 1 if a going concern opinion is issued in the year, 0 otherwise

ε

The residual of the charged audit fee

3.2 Measuring earnings management

One of those indirect methods is the level of earnings management by the auditee and the extent to which this is tolerated by the auditor. According to Becker et al. (1998), the level of earnings management that an auditor tolerates is an indirect proxy for audit quality. If an auditor is independent, he would not tolerate earnings management and thus constrain earnings management.

(17)

The compensation of managers is commonly based on the reported earnings. According to the agency theory, managers are trying to maximize their own wealth. So they act in their personal interest (Jensen and Meckling, 1976). Managers have the incentive to manipulate the reported earnings in order to maximize their compensation.

Discretionary accruals show the consequences of discretionary actions that managers have made to influence their short-term earnings. In recent prior research discretionary accruals are measured using the modified model of Jones. Whereby total accruals (TA) will be computed as follows:

-$A = ∆5$ − ∆5! − ∆5DEℎ + ∆G5! − GHI $AJ,

Secondly, there follows an OLS regression analysis to show firm specific industry attributes. The beta (b) estimates the relationship between the independent variable and the dependent variable. In this case the independent variables are characteristics of the relationship between auditor and auditee. The dependent variables are the proxies for audit quality.

When the betas are estimated the following model will be used to estimate the non-discretionary accruals. "G$A = )*+ ), 1 $AJ,+ ). ∆0&4A− ∆0&5A $AJ, + )L ??&A $AJ, + @A

Discretionary accruals are the difference between total accruals and the non-discretionary accruals. So to compute the discretionary accruals, the non-discretionary accruals need to be subtract from the total accruals.

(18)

Variable Description

TA Total accruals

At-1 Lagged Assets

DCL Change in current liabilities

DCA Change in current assets

DCash Change in cash

DDCL Change in debt in current liabilities Dep Depreciation

NDA Non-discretionary accruals DA Discretionary accruals

DREV Change in revenues

DREC Change in receivables

PPE Property, plant and equipment

3.3 Measuring investor protection

The degree of investor protection can be assessed in multiple ways. The most common variable used to define investor protection in prior literature are the variables of La Porta et al. (1998). In their research, they examined the legal rules that cover the protection of shareholders and creditors in 49 countries. They developed a variable that scores the investor protection per country. They distinguish multiple components that can measure the level of investor protection in a country. To measure investor protection

In the paper of La Porta et al. (1998) two types of laws were distinguished: the common law and the civil law. The common law arises from the traditional English law. This law is constructed by preceding judgments that are useful to solve unambiguous cases. The civil law arises originally from the Roman times. This law is based on core principles that are applicable to solve different processes. Hereby, preceding judgments play no role in these proceedings. According to La Porta (1998), the level of investor protection is the highest in common-law countries. In these countries the auditor will be sooner exposed to auditor liability. Besides, La Porta et al. (1998) concentrated on the corporate law of a nation by building up the counter chief rights' list. This file depends on six unique components of financial specialist security in a nation's corporate law. The file measures how investors can exercise their rights towards the executive board of an organization. This score is called the anti-director’s index.

(19)

Furthermore, the role of the public enforcer is being rated. This represents the rate of allowance of corruption by a country’s law. Appendix 1 shows the composition of the scores used in the empirical model.

3.4 Sample selection

The data will be retrieved from two different databases: Compustat and Audit Analytics over a period of 2010-2015. This data will be merged through linking the Company Identification Key (CIK). The sample includes global data from the 49 countries for which the level of investor protection is scored in the paper of La Porta et al. (1998). The initial sample consists of 10.208 firm-year observations. Data from the financial sector will be excluded for

comparability reasons. The SIC codes 6000-6999 were excluded from the Compustat sample and thus will not be included in the final sample. Subsequently, all incomplete financial data will be excluded. The final sample exists of 6255 firm-year observations from 2284 different firms. From which 2871 firm year observations are non-US. Only 1064 firm year

observations come from countries below the mean and are associated with weak investor protection1.

Table 3.1 Formation of the sample Firm year observations Description

44.045 Initial sample

25.576 -/- merging

12.214 -/- Incomplete data

6.255 Final sample

3.5 Empirical model

Beside the hypothesized relationship between the level of discretionary accruals and abnormal audit fees and the level of investor protection, prior literature shows multiple variables that influence the level of discretionary accruals. Therefore, several variables will be included in the

(20)

Client size can have a positive influence on the quality of financial reporting. Because larger firms have more advanced internal control systems, which results in less opportunities for managers to engage in earnings management (Becker et al., 1998). Furthermore, large firms tend to have a lower level of discretionary accruals, because of more stable and predictable operations. According to Frankel et al. (2002) companies with high cash flows of operations (CFO) are associated with a lower level of earnings management. The variables LEV, LOSS, and LIQ control for the financial position of a company. Because companies with financial difficulties are more sensitive to engage in earnings management (Ashbaugh et al., 2003). Furthermore, prior literature argues that non-audit fees have a negative impact on the level of earnings management (Frankel et al., 2002 and Ashbaugh et al., 2003). Non audit fees increases the economic bond and are a threat to auditor independence. The natural logarithm of the non-audit fees is taken, because the natural logarithm minimizes the effect of greater values with regard to the effect of smaller values.

To test the hypotheses, the following model is designed:

$=;_G$ = )* + ),$=;_$="%&& + ). 1"4?0/ + )L$="%&& ∗ 1"4?0/

+ )3!"-$ + )65%/ + )70/$ + )8!&4 + ):!19 + )<!/;; + ),*=124 + ),,!""$% + @

Table 3.3 Definitions of the empirical model

Variable Description

ABS_DA ABS_ABNFEE

The absolute value of discretionary accruals The absolute value of the abnormal audit fee

INVPRO The investor protection scores according to La Porta et al. (1998) LNTA Natural logarithm of total assets

CFO Cash flow from operations divided total assets ROA Net income divided by total assets

LEV LIQ

Total debt of the firm divided by the total equity Current assets divided by current liabilities

LOSS 1 if the firm reported a loss in the prior year, otherwise0 BIG4 1 if the firm has a Big 4 auditor, otherwise 0

LNAF Natural logarithm of non-audit fees

(21)

4

Analyses and Results

In this paragraph the descriptive statistics and correlation of the variables will be discussed. Subsequently, the regression results of the expected audit fee model and the empirical model will be analyzed. Finally, an additional analysis will be performed on two subsamples of discretionary accruals.

4.1 Descriptive statistics

Along with the expected audit fee model described in paragraph 3.2 we calculated the abnormal audit fees upon which will be relied in the empirical model. I refer to table 4.1 for the descriptive statistics of all variables in the expected audit fee model. To prevent that outliers may significantly impact the final results, the continuous data has been winsorized. Winsorizing means that the values of the outliers are being replaced by less extreme values, to achieve a more evenly distributed sample. The following table summarizes the descriptive statistics of the variables used in the expected audit fee model, as well as the empirical model.

Table 4.1 Descriptive statistics

Variable Mean Std. dev. Minimum Maximum

ABS_DA 0.0492 0.7652 9.55e-6 31.6531 LNAFEE 13.4532 1.7135 9.8601 16.1296 ABNFEE 0.0489 0.9057 -6.0809 3.3831 INVPRO 9.1937 0.8937 4.084 9.992 LNTA 6.2333 2.5037 1.1727 10.4039 CFO 0.0231 1.6199 -3.7634 160.1931 FOREIGN 0.0249 2.6995 -66.6666 170.8182 INVREC 0.2119 0.1661 -66.6666 0.9029 ROA -0.0823 0.6401 -10.0648 47.4091 LEV 0.3567 4.8135 0 27.4889 LIQ 2.8066 3.5603 0.0303 142.7286

(22)

OPINION 0.0681 0.2519 0 1

LNNAF 11.8626 1.841501 1.7917 15.0393

This table displays the descriptive statistics of the variables in the expected audit fee model and the empirical model. This includes both the dependent, independent, moderating and control variables.

Table 4.2 shows the descriptive statistics of the variables used in the empirical model. The variable ABS_DA represents the absolute value of the discretionary accruals, which is used as a proxy for earnings management. The mean of 0.0492 is in line with prior research (Choi et al. 2010, Hoitash et al., 2007 and Kraub et al., 2015). The mean of ABNFEE is … and is in line with prior research of Choi et al. (2010). The variable INVPRO is the level of investor protection scored according to La Porta (1998). The mean of the score is 9.1937 and is close to the maximum score of 9.992. This means that a major part of the sample comes from countries with high investor protection. Notable is that the mean of FOREIGN is quite low, this indicates that the firms in the sample are on average not really active outside their country. The mean of LEV of 0.3567 suggests that most firms are more equity financed. The mean of investor protection The mean of LOSS shows that 40.14% of the firms reported a loss in the prior year. Furthermore, 77.62% of all firm-year observations is audited by a Big 4 auditor. In only 6.81% of the firm-year observations a firm received a going concern opinion.

4.2 Correlation

To check for any correlation between the different variables of the empirical model. In appendix 2 the pairwise correlation between the variables is displayed. In addition, this correlation matrix gives a general indication of the relation between the variables. The variables ABS_ABNFEE, INVPRO, ROA, LEV, LOSS, BIG4 and LNNAF all appear to be significantly correlated with the absolute discretionary accruals. The signs of their coefficients seem consistent with the expectations of this study, except for LNNAF. Moreover, it is worth mentioning that there is a significant correlation between the variables LNTA and BIG 4 suggests that firms with larger assets are more likely to hire a Big 4 auditor. Also the correlation between LNTA and LOSS is negatively significant, suggesting that firms with larger assets are less likely to report a loss. Furthermore, the significant correlation between LNTA and LNNAF. Firms with larger assets have more non-audit fees. The Pearson

(23)

cohesion between the variables of interest and thus, it is not probable that those variables measure the same phenomena.

4.3 Results of the estimation of abnormal audit fees

First the abnormal audit fees are estimated as the error terms of the expected audit fee model. In table 4.2 below, the regression results of this model are shown. All variables with

significant coefficients (LNTA, INVREC, LOSS, BIG4, OPINION) are consistent with the predicted direction. The coefficients of the variables are ROA and LEV are not in line with the predicted directions, but therefore not significant. The adjusted R2 indicates that 69.06% of the total variance of the audit fees is explained by thee model. Prior research of Hoitash et al. (2007) and Blankley et al. (2012) measured an R2 of 0.728 and 0.790, respectively.

Although the adjusted R2 of this regression is lower than prior research, the expected audit fee model still explains most of the variance, indicating that the abnormal audit fees are estimated appropriately.

Table 4.2 Results of regression on expected audit fee model

Variable Expected sign Coefficient (b) Std. err. (b) t-value Significance

LNTA + 0.5794*** 0.0053 109.4 0.000 FOREIGN + 0.0026 0.0036 0.77 0.480 INVREC + 1.6013*** 0.0623 6.44 0.000 ROA + -0.004 0.0109 -0.03 0.972 LIQ - -0.0057 0.0036 -1.58 0.115 LEV + -0.0094 0.0246 -0.38 0.703 LOSS + 0.2532*** 0.0227 11.13 0.000 BIG4 + 0.0988*** 0.0283 3.94 0.000 OPINION + 0.1988*** 0.0504 3.49 0.000 Constant 9.3411*** 0.0461 202.74 0.000

* significant at level p=0.10 Observations = 6255 ** significant at level p=0.05 Adjusted R2 = 0.6906

(24)

4.4

Main analysis

In order to confirm or reject the three hypotheses, multiple regressions are performed. The first regression measures the relationship between the absolute discretionary accruals and the independent variables in the empirical model, based on the total sample. The results of this regression are displayed in table 4.3. On the basis of these results hypothesis 2 and 3 can be confirmed or rejected. With regard to the first hypotheses an overall conclusion can be drawn from this regression. Hereafter there will follow a more detailed regression where the total sample will be divided into a subsample containing positive abnormal audit fees and a subsample containing negative abnormal audit fees. This model is significant with p<0.01 and has an adjusted R2

0.0455. Meaning that 4.55% of the total variance in the absolute discretionary accruals is explained by the independent variables of this regression model. Striking is that the relationship between absolute abnormal audit fees (ABS_ABNFEE) and the absolute discretionary accruals (ABS_DA) is significantly negative at the level p<0.01. This result is in contrast to the expectations that both positive and negative abnormal audit fees would be positively related to discretionary accruals. The combined hypothesis 1, that abnormal audit fees would increase earnings management, can be rejected by the results of this first regression. The second, more detailed, regression analysis will gain further insight in the differences between positive and negative abnormal audit fees.

The negative coefficient of the investor protection variable (INVPRO) indicates that the level of investor protection reduces the level of absolute discretionary accruals. However, this coefficient is not found to be significant. Hence, this relationship cannot be confirmed, the second hypothesis is rejected. The interaction variable ABNFEE*INVPRO measures the influence that the level of investor protection has on the relationship between abnormal audit fees and discretionary accruals. Despite the insignificant relation of investor protection variable on the absolute discretionary accruals, the results show a significant negative coefficient for the interaction variable. Thus, hypothesis 3 can be confirmed based by the results on the total sample. Furthermore, the coefficients of the control variables LNTA, ROA, LEV, LIQ and LOSS are found to be significant in the predicted directions.

From this regression can be concluded that abnormal audit fees have do not negatively affect the level of earnings management. Secondly, the level of investor protection does not influence the level of earnings management. Lastly, the significant negative coefficient of the interaction variable shows that investor protection mitigates the relation between abnormal

(25)

audit fees and earnings management. These conclusions are drawn with regard to the total sample.

Table 4.3 Results of regression on the empirical model (total sample) Dependent variable = ABS_DA

Independent Variable Expected sign Coefficient (b) Std. err. (b) t-value Significance

ABS_ABNFEE + -0.1807*** 0.0686 -2.64 0.008 INVPRO - -0.0183 0.0144 -1.26 0.206 ABNFEE*INVPRO - -0.0341*** 0.0083 4.09 0.000 LNTA - -0.0134** 0.0059 2.27 0.023 CFO - 0.6810*** 0.0732 9.31 0.000 ROA - -0.3748*** 0.0494 -7.59 0.000 LEV + 0.0191*** 0.0334 7.56 0.000 LIQ - -0.0104*** 0.0036 -2.87 0.004 LOSS - -0.0686*** 0.0059 -4.16 0.000 BIG4 - -0.2441 0.0258 -0.94 0.345 LNNAF + -0.0245*** 0.0196 3.50 0.000 Constant -0.6881*** 0.1387 4.96 0.000

* significant at level p=0.10 Obs = 6255

** significant at level p=0.05 Adjusted R2 = 0.0455

*** significant at level p=0.01 Prob > F = 0.000

Although the expectations of this study, with regard to the impact of positive and negative abnormal audit fees on the value of absolute discretionary accruals, were similar. The results of the first regression show that the influence on discretionary accruals is different. So in order to distinguish different effects of positive or negative abnormal audit fees on discretionary accruals, the total sample will be divided. This is necessary to properly answer hypothesis 1a and 1b. Two separate regressions of the empirical model are performed, one with the subsample of positive abnormal audit fees and one The results of this regression are shown in table 4.4.

(26)

accruals is explained by the variables in this regression model. The significant negative coefficient of the beta of ABNFEE is in accordance with the findings of Eshleman & Guo (2014). Their study follows the effort theory of Hay et al. (2006) and explain this result as an increase of audit effort, to the extent that the positive abnormal audit fee increases. Thus, lowering the risk that earnings management would not be detected. Therefore, hypothesis 1a can be rejected. Engagements with positive abnormal audit fees have a positive influence on earnings management.

With reference to the level of investor protection (INVPRO) and absolute discretionary accruals, no significant relation is found. The sign of this coefficient is positive, this would suggest that the level of investor protection should increase the level of the absolute discretionary accruals. But, because this coefficient is insignificant, no conclusion can be drawn upon the relationship between the level of investor protection and earnings management. Hypothesis 2 cannot be confirmed for the subsample of positive abnormal audit fees. Concerning the interaction variable ABNFEE*INVPRO, the results show a significant negative coefficient of beta. This is in accordance with the expectations. The level of investor protection can increase the risk of auditor liability and legal exposure. This could mitigate the incentive that auditors have to report favorably for their clients due to positive abnormal audit fees. The results of this regression suggest that level of investor protection has a negative effect on the relationship between positive abnormal audit fees and absolute discretionary accruals. Therefore, hypothesis 3 can be confirmed for this subsample.

With regard to the subsample with negative abnormal audit fees, the explanatory power of the independent variables is 5.75%. This subsample contains 2503 firm year observations. The regression model of the subsample is significant at a level of p<0.01. The variable ABNFEE shows a positive but insignificant coefficient, with regard to absolute discretionary accruals. Indicating that negative abnormal audit fees are not related to absolute discretionary accruals. This is consistent with the finding of Choi et al. (2010). Likewise, they did not find a relationship between negative abnormal audit fees and discretionary accruals. The significant positive relation between negative abnormal audit fees and the absolute value of the discretionary accruals rejects hypothesis 1b. Negative abnormal audit fees do not influence the level of discretionary accruals.

The INVPRO variable indicates that there is a negative relation between the level of investor protection and the absolute value of discretionary accruals. This relationship in line with hypothesis 2: the level of investor protection has a negative relationship with the level of

(27)

discretionary accruals. However, the coefficient is not found to be significant. Hence, the second hypothesis cannot be confirmed concerning negative abnormal audit fees.

The interaction variable ABNFEE*INVPRO is insignificant. The level of investor protection does not impact the relationship between negative abnormal audit fees and the absolute discretionary accruals. The results of this regression in table 4.4 show a significant negative coefficient for this interaction variable. This is in accordance with the expectation that the level of investor protection mitigates the effect of abnormal audit fees on discretionary accruals.

Furthermore, the following control variables in this regression have a significant relationship with the absolute discretionary accruals: CFO, ROA, LEV, LIQ, BIG4 and LNNAF. However, the directions of the coefficients of CFO and LNNAF are not consistent with the expectations. Meaning that in the subsample of negative abnormal audit fees the natural logarithm of the non-audit fees decreases the absolute value of the discretionary accruals. Hence, in this subsample non-audit fees reduces earnings management.

Table 4.4 Regression results of divided sample of abnormal audit fees

Dependent variable = ABS_DA Subsample

Independent variable Expected sign Positive ABNFEE Negative ABNFEE ABNFEE1 + -0.6444** (0.011) 0.3675 (0.898) INVPRO - 0.0472 (0.169) -0.0239 (0.308) ABNFEE*INVPRO - -0.0682** (0.011) -0.0640 (0.172) LNTA - -0.0167*** (0.001) 0.0222 (0.073) CFO - -0.3292*** (0.000) 1.6267*** (0.000) ROA - -0.1864*** (0.000) -0.9151*** (0.000)

(28)

LIQ - -0.0109*** (0.001) -0.0113* (0.097) LOSS + -0.0441*** (0.006) 0.0714 (0.106) BIG4 - -0.0512** (0.016) -0.0070*** (0.898) LNNAF + -0.0039 (0.438) -0.0142* (0.297) Constant -0.1290 (0.608) 0.04767*** (0.036) * significant at level p=0.10 ** significant at level p=0.05 *** significant at level p=0.01 Obs = 3752 Adj. R2 = 0.0708 Prob>F= 0.0000 Obs = 2503 Adj. R2 = 0.0575 Prob>F= 0.0000

1 The variable ABNFEE is represented by the subsample with positive abnormal audit fees or negative abnormal audit fees.

4.5 Additional analysis

In the first regression (table 4.3) the results show a significant negative relationship between absolute audit fees and absolute discretionary accruals. The second regression shows a significant negative relation between positive abnormal audit fees and the absolute

discretionary accruals. However, the coefficient of ABNFEE, does not show a relationship between negative abnormal audit fees and absolute discretionary accruals. To create a further understanding of the impact of abnormal audit fees and investor protection on discretionary accruals, an additional analysis will be performed.

According to Burgstahler & Dichev (1997), there are two forms to manage earnings, namely managing earnings upwards or downwards. Therefore, the discretionary accruals will be divided into two subsamples, namely positive discretionary accruals and negative

discretionary accruals. Those are proxies for income increasing earnings management and income decreasing earnings management, respectively. For the negative discretionary accruals subsample, the inverse of the predicted signs is expected. The distribution of the subsamples is displayed in table 4.7.

(29)

The relationship of positive and negative abnormal audit fees on positive discretionary accruals is displayed in table 4.5. It shows that the subsample of positive abnormal audit fees contains 722 observations. This regression model explains 11.01% of the total variance in the positive discretionary accruals. The model is significant at p<0.01.

The coefficient of ABNFEE is positive and significant in this subsample. Indicating that there is a relationship between positive abnormal audit fees and positive discretionary accruals. Despite, the results do not show a significant relationship between the level of investor protection and the level of positive discretionary accruals. Nonetheless, the coefficient of the interaction variable ABNFEE*INVPRO is negatively significant. Indicating that the level of investor protection has a negative influences the relationship between positive abnormal audit fees and positive discretionary accruals. Hence, positive abnormal audit fees increase income increasing earnings management. Besides, investor protection has a mitigating effect on the relationship between positive abnormal audit fees and income increasing earnings management. The control variables LNTA, LEV and LOSS are also significant. These variables all have significant relationship in the predicted directions. Notable is that, unlike the significant negative coefficient for ABNFEE of positive abnormal audit fees (table 4.4), the results now show a positive sign. Supporting the economic dependence theory of DeAngelo (1981a) for positive discretionary accruals. Hence, hypothesis 1a can be confirmed with regard to income increasing earnings management.

According to table 4.5, the subsample of negative abnormal audit fees consists of 360 observations. The adjusted R2

is 0.1555, meaning that the total variance in positive discretionary accruals is explained for 15.5% by the variables in the regression model. The model is significant at the level p<0.01. The variables of interest do not show significant coefficients with regard to positive abnormal audit fees. Suggesting that there is no relation between ABNFEE, INVPRO or ABNFEE*INVPRO and income increasing earnings management in the subsample with negative abnormal audit fees. However, the control variables LNTA and LOSS are significant in the predicted directions. The variables of interest not showing a significant coefficient, is in line with the results found for the regression of the total discretionary accruals sample.

(30)

Table 4.5 Regression results of the ABNFEE subsamples concerning positive discretionary accruals

Dependent variable= POS_DA Subsample

Independent variable Positive ABNFEE Negative ABNFEE ABNFEE1 3.2046** (0.015) 2.2685 (0.681) INVPRO -0.1548 (0.133) -0.2925 (0.175) ABNFEE*INVPRO -0.3293** (0.019) -0.3767 (0.167) LNTA -0.0488*** (0.007) -0.1964*** (0.003) CFO 0.1671 (0.100) -0.7741 (0.317) ROA -0.0152 (0.802) -0.0282 (0.961) LEV 0.4352*** (0.000) 0.3491 (0.423) LIQ 0.0095 (0.253) -0.0054 (0.826) LOSS 0.1344** (0.024) 0.0873* (0.099) BIG4 -0.0277 (0.152) -0.3101 (0.171) LNNAF -0.0201 (0.235) -0.1038 (0.694) Constant -1.3865 (0.152) 2.6224** (0.057) * significant at level p=0.10 Obs = 722 Obs = 360 ** significant at level p=0.05 Adj. R2 = 0.1101 Adj. R2 = 0.1555 *** significant at level p=0.01 prob>F= 0.0000 prob>F= 0.0000

1 The variable ABNFEE is represented by the subsample with positive abnormal audit fees or

(31)

Table 4.6 show the results of the regression concerning negative discretionary accruals. The positive abnormal audit fees subsample contains 3390 firm year observations. The model is significant at the level p<0.01 and has an adjusted R2 of 0.0716. The coefficient of ABNFEE is positive, but is only significant at p<0.1. Indicating that there is a positive relation between excessive audit fees and negative discretionary accruals. Hence, overpayment of the auditor lead to less income decreasing earnings management. This result supports the effort theory of Hay et al. (2013) and is consistent with the findings in table 4.4. The other variables of interest INVPRO and ABNFEE*INVPRO do not show a significant coefficient. Suggesting no relationship between these variables and income decreasing earnings management. However, the control variables LNTA, CFO, ROA, LIQ LOSS and BIG4 all appear to be significant.

The regression results of the negative abnormal audit fees subsample the model is significant at p<0.01. The sample contains 1783 observations. The total variance of negative discretionary accruals is explained by the variables in the regression model. Regarding negative abnormal audit fees and negative discretionary accruals, there seems to be no relationship. This result indicates that the underpayment of the audit engagement does not lead to a change in income decreasing earnings management. Also, the level of investor protection (INVPRO), nor the interaction term ABNFEE*INVPRO would change the level of income decreasing earnings management. Only the control variables CFO, ROA, LEV, LIQ, LOSS, BIG4 and LNNAF are found to be significant, hence influential.

Table 4.6 Regression results of the ABNFEE subsamples concerning negative discretionary accruals

Dependent variable= NEG_DA1 Subsample

Independent variable Positive ABNFEE Negative ABNFEE ABNFEE2 0.4789* (0.057) -0.0072 (0.923) INVPRO -0.0366 (0.167) -0.0265 (0.117) ABNFEE*INVPRO 0.0527 (0.175) 0.0037 (0.695) LNTA -0.0161*** -0.0038

(32)

ROA -0.2779*** (0.000) 1.2403*** (0.000) LEV 0.2457*** (0.000) -0.1502*** (0.022) LIQ 0.0132*** (0.000) 0.0191*** (0.001) LOSS -0.0255*** (0.124) -0.0693** (0.036) BIG4 0.0583*** (0.008) -0.0975** (0.022) LNNAF 0.0722 (0.173) 0.0222** (0.028) Constant -0.3069 (0.234) -0.2492 (0.145)

* significant at level p=0.10 Obs = 3390 Obs = 1783 ** significant at level p=0.05 Adj. R2 = 0.0716 Adj. R2 = 0.1125 *** significant at level p=0.01 prob>F= 0.0000 prob>F= 0.0000

1 The value of the negative discretionary accruals is not absolute

2The variable ABNFEE is represented by the subsample with positive abnormal audit fees or

negative abnormal audit fees.

From this additional analysis can be concluded that negative abnormal audit fees neither have an effect on positive discretionary accruals, nor negative discretionary accruals. Suggesting that underpayment of the auditor does not influence earnings management. Thus, hypothesis 1b can be rejected. This finding is in line with the research of Choi et al. (2010). In contrast to this, the finding concerning positive abnormal audit fees are significant in both ways. Positive abnormal audit fees significantly increase income increasing earnings management. This finding is consistent with the theory of economic dependence (DeAngelo, 1981a). However, income decreasing earnings management decreases by the influence of positive abnormal audit fees. Furthermore, none of the regressions show a significant coefficient for investor protection. Indicating that there is no relationship between. This finding is consistent with the results of the regressions included in table 4.3 and 4.4. Therefore, the second hypothesis can be rejected. With regard to the third hypothesis, the results indicate that the mitigating effect of investor protection is only significant on the relationship between positive abnormal audit

(33)

Table 4.7 Descriptive statistics of subsamples

Variable Mean Std. dev. Minimum Maximum

POS_ABNFEE 0.6092 0.4485 0.0019 3.3831

NEG_ABNFEE -0.7829 0.9389 -6.0809 -0.0005

POS_DA 0.1359 1.4055 9.55e-6

31.6531

(34)

5 Conclusion

5.1 Summary

After different accounting scandals as Enron and WorldCom, audit quality and auditor independence were being questioned. There was argued that auditor-client relations, were there was a strong economic bond, could lead to comprised auditor independence (DeAngelo, 1981a and Simunic, 1984). A vast amount of literature examined the relationship between different measures of economic dependence and audit quality.

This study attempted to answer the following question: does investor protection mitigates the relation between abnormal audit fees and earnings management? With abnormal audit fees being a proxy for economic dependence and earnings management is used as a proxy for audit quality. The level of earnings management is estimated by the Modified Jones Model.

According to prior literature there are two different theories with regard to abnormal audit fees and earnings management. The economic dependence theory argues that a stronger economic bond comprises auditor independence (DeAngelo, 1981a and Simunic, 1984). Auditors are inclined to report more favorable and thus have a greater allowance towards earnings management. On the opposite, there is the effort theory of Hay et al. (2006). They argue that excessive fees increase the audit effort and thus reduce earnings management. Regarding investor protection, it is argued that the more an auditor is exposed to legal liability, the less allowance they have towards earnings management. Because auditors face litigation costs. The predictions of this study are based on prior literature. The following relations were hypothesized:

1. Both positive and negative abnormal audit fees have a negative influence on earnings management

2. The level of investor protection positively influences earnings management

3. The level of investor protection positively influences the relationship between abnormal audit fees and earnings management

Several regressions were performed to test the hypothesized relations. The first regression is based on the total sample. The second regression distinguishes positive from negative abnormal audit fees regarding the value of absolute discretionary accruals. In addition, an analysis will

(35)

discretionary accruals. These subgroups represent income increasing earnings management and income decreasing earnings management, respectively.

The results show to types of relations with regard to positive abnormal audit fees and earnings management. First, the overpayment of an audit engagement has a mitigating effect in case of income decreasing earnings management. Second, these excessive fees have an increasing effect on income increasing earnings management. Furthermore, hypothesis 1b can be rejected. The findings are not showing any significant relation between the underpayment of an audit engagement and the level of earnings management. The results do not show a significant coefficient for the variable of investor protection in any of the regressions performed. Indicating that there is no relationship between investor protection and earnings management. Hence, the second hypothesis is rejected. For the hypothesized mitigating effect of investor protection on the relationship between abnormal audit fees and discretionary accruals, evidence is only found in case of positive abnormal audit fees and positive discretionary accruals.

5.2 Conclusion

This research indicates that there is a difference in effect between positive abnormal audit fees and two forms of earnings management. In case of managing earnings upwards, the results show that positive abnormal audit fees ensure an increase in positive discretionary accruals. This result is supported by the economic dependence theory of DeAngelo (1981a). However, positive abnormal audit fees mitigate the level of discretionary accruals in case of income decreasing earnings management. This finding is in accordance with the effort theory of Hay et al. (2006). Additionally, the results argue that negative abnormal audit fees do not influence the level of earnings management. Furthermore, the findings show no evidence for a significant relationship between the level of investor protection of a country and the level of earnings management. This can be interpreted as investor protection standalone does not influence earnings management. The results of this study only find evidence for the mitigating effect of investor protection on the relationship between positive abnormal audit fees and income increasing earnings management.

(36)

countries. The insignificant relationship between investor protection and the level of earnings management, could be due to this limitation. Furthermore, different industries are included in this sample. Each type of industry may have different incentives with regard to earnings management. Moreover, the estimation model of the expected audit fee model had an adjusted R2 of 0.6906. The abnormal audit fees are estimated as the error terms of this model. Because the model is not fully explanatory, this could lead to bias and estimation errors of the abnormal audit fees, which cannot be ignored.

The level of discretionary accruals is only a proxy for audit quality, but to fully capture audit quality. Future research might use more proxies for audit quality. Furthermore, future research might redo this research with another variable that indicates the economic dependence, like non-audit fees or client importance (Frankel et al., 2002 and Francis & Reynolds, 2000)

Referenties

GERELATEERDE DOCUMENTEN

Deze Big Data Revolutie wordt ook uitmuntend beschreven in het boek ‘De Big Data Revolutie’, waarin big data wordt beschreven als bron van economische waarde en

affordable, reliable, clean, high-quality, safe and benign energy services to support economic and human

The expanded cells were compared with their unsorted parental cells in terms of proliferation (DNA content on days 2, 4, and 6 in proliferation medium), CFU ability (day 10

Het verschil in effect tussen live en uitgesteld kijken op merkbekendheid en merkreputatie van de sponsor kan mogelijk worden gemedieerd door de mate van betrokkenheid bij

This chapter described the running-in of rolling-sliding contacts on macroscopic and microscopic level. 1) On macro-scale, the geometrical change of the contacting

Tiago Filipe Montes de Jesus University of

ten Elshof, Electronic band structure and electron transfer properties of two- dimensional metal oxide nanosheets and nanosheet films, Curr. Solid State

Recoin is a recommender system for understanding and improving the completeness of entities in Wikidata [1, 3]. A main motivation for implementing Recoin is Wikidata’s openess, since